Provided is a robot device including an image input unit for inputting an image of surroundings, a target object detection unit for detecting an object from the input image, an object position detection unit for detecting a position of the object, an environment information acquisition unit for acquiring surrounding environment information of the position of the object, an optimum posture acquisition unit for acquiring an optimum posture corresponding to the surrounding environment information for the object, an object posture detection unit for detecting a current posture of the object from the input image, an object posture comparison unit for comparing the current posture of the object to the optimum posture of the object, and an object posture correction unit for correcting the posture of the object when the object posture comparison unit determines that there is a predetermined difference or more between the current posture and the optimum posture.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A mobile apparatus, comprising: circuitry configured to: input environment information regarding an environment of the mobile apparatus including an object and posture data of the object, including a set of coordinates of the object in the environment, to a learning model that has learned to output target posture data regarding a target posture of the object based on the inputted environment information and the posture data of the object, the learning model repeating observations of the object to obtain a plurality of sets of the coordinates of the object using the environment information and the posture data until a total number of the observations reaches a predetermined number; in response to the total number of the observations reaching the predetermined number, acquire the target posture data of the object from the learning model, by setting, as the target posture data, the set of the coordinates of the object having been observed a most number of times through the repeating of the observations of the object; acquire an image of the environment including the object; obtain current posture data regarding a current posture of the object based on the acquired image of the environment, the current posture data including the set of the coordinates of the object; and perform a process of correcting the posture of the object based on the target posture data and the current posture data by comparing the set of the coordinates of the object for the target posture data and the current posture data, such that the current posture of the object becomes closer to the target posture.
2. The mobile apparatus according to claim 1 , wherein the circuitry is further configured to: acquire the environment information including the position of the object; and input the environment information on to the learning model.
3. The mobile apparatus according to claim 2 , wherein the circuitry is further configured to: detect the position of the object from the image; acquire the environment information including the position of the object; and input the environment information to the learning model.
4. The mobile apparatus according to claim 1 , wherein the learning model is updated to output the target posture data based on a frequency of combination of the environment information and the posture data of the object.
5. The mobile apparatus according to claim 1 , wherein the learning model is updated to output the target posture data based on user input.
6. The mobile apparatus according to claim 1 , wherein the circuitry is further configured to: obtain posture change data indicating an amount of a change of the posture of the object to be corrected in a case that a predetermined difference exists between the target posture data and the current posture data; and perform the process of correcting the posture of the object based on the posture change data.
7. The mobile apparatus according to claim 6 , wherein the circuitry is further configured to: cause an end effector to move as the process of correcting the posture of the object based on the posture change data.
8. A method comprising: inputting, by circuitry of a mobile apparatus, environment information regarding an environment of the mobile apparatus including an object and posture data of the object, including a set of coordinates of the object in the environment, to a learning model that has learned to output target posture data regarding a target posture of the object based on the inputted environment information and the posture data of the object, the learning model repeating observations of the object to obtain a plurality of sets of the coordinates of the object using the environment information and the posture data until a total number of the observations reaches a predetermined number; in response to the total number of the observations reaching the predetermined number, acquiring, by the circuitry, the target posture data of the object from the learning model, by setting, as the target posture data, the set of the coordinates of the object having been observed a most number of times through the repeating of the observations of the object; acquiring an image of the environment including the object; obtaining, by the circuitry, current posture data regarding a current posture of the object based on the acquired image of the environment, the current posture data including the set of the coordinates of the object; and performing, by the circuitry, a process of correcting the posture of the object based on the target posture data and the current posture data by comparing the set of the coordinates of the object fax the target posture data and the current posture data, such that the current posture of the object becomes closer to the target posture.
9. The method of claim 8 , further comprising: acquiring the environment information including the position of the object; and inputting the environment information to the learning model.
10. The method of claim 9 , further comprising: detecting the position of the object from the image; acquiring the environment information including the position of the object; and inputting the environment information to the learning model.
11. The method of claim 8 , wherein the learning model is updated to output the target posture data based on a frequency of combination of the environment information and the posture data of the object.
12. The method of claim 8 , wherein the learning model is updated to output the target posture data based on user input.
13. The method of claim 8 , further comprising: obtaining posture change data indicating an amount of a change of the posture of the object to be corrected in a case that a predetermined difference exists between the target posture data and the current posture data; and performing the process of correcting the posture of the object based on the posture change data.
14. The method of claim 13 , further comprising: causing an end effector to move as the process of correcting the posture of the object based on the posture change data.
15. The mobile apparatus according to claim 1 , wherein the circuitry is configured to se a posture learned flag indicating an optimum posture can be calculated by sufficiently observing the object in the environment.
16. The mobile apparatus according to claim 15 , wherein the circuitry is configured to set the posture learned flag in response to determining that a total number of observation of the environment has reached the predetermined number.
17. The mobile apparatus according to claim 1 , wherein the circuitry is configured to: determine whether a predetermined difference or more exists between the target posture data and the current posture data; and perform the process of correcting the posture of the object in a case where the predetermined difference or more is determined to exist between the target posture data and the current posture data.
18. The method of claim 8 , further comprising: determining, using the circuitry, whether a predetermined difference or more exists between the target posture data d the current posture data; and performing, using the circuitry, the process of correcting the posture of the object in a case where the predetermined difference or more is determined to exist between the target posture data and the current posture data.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 23, 2019
May 31, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.